PINE BLUFF ARSENAL D BY PINE BLUFF, ARKANSAS TECHNICAL REPORT QAL * COPY

Transcription

1 -* COPY PINE BLUFF ARSENAL PINE BLUFF, ARKANSAS TECHNICAL REPORT QAL-89-1 HEXACHLOROETHANE PURITY AND ASSAY IN SMOKE MIX BY HIGH PRESSURE LIQUID CHROMATOGRAPHY DTIC MT TASK # 1705 ELECTE 0s- f U MAR D BY RONALD WISE CARL HENRY CAROLYN BRANSCOMB TIMOTHY McGAULEY RON HOFFMAN, DWI--I)uy ON STA r A PATRICK BROWN Approved for public,.leas STEMEN LOWREY D.,. n~jtuon Urndi;ted DECEMBER 1989

2 SECURITY CLASSIFICATION OF THIS PAGE REPORT DOCUMENTATION PAGE O MB No la. REPORT SECURITY CLASSIFICATION lb. RESTRICTIVE MARKINGS UNCLASSIFIED 2a. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION /AVAILABILITY OF REPORT Approved for Public Release; 2b. DECLASSIFICATION /DOWNGRADING SCHEDULE distribution is unlimited. 4. PERFORMING ORGANIZATION REPORT NUMBER(S) S. MONITORING ORGANIZATION REPORT NUMBER(S) PBA-TR-QAL a. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a. NAME OF MONITORING ORGANIZATION (if applicable) Pine Bluff Arsenal SMCPB-QAL 6c. ADDRESS (City, State, and ZIPCode) 7b. ADDRESS (City, St=o, and ZIP Code) Pine Bluff, AR a. NAME OF FUNDING/SPONSORING 8b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER ORGANIZATION (If applicable) Materials Technology Laboratory SLCMT-MSI-QA 8c. ADDRESS (City, State, and ZIP Code) 10. SOURCE OF FUNDING NUMBERS PROGRAM PROJECT TASK WORK UNIT Watertown, MA ELEMENT NO. NO. NO. ACCESSION NO. MTT-M88 Y6780=I1E TITLE (Include Security Classification) Hexachloroethane Purity and Assay in Smoke Mix by High Pressure Liquid Chromatography 12. PERSONAL AUTHOR(S) Ronald Wise; Carl Henry; Carolyn Braiscomb; Timothy McGauley, Ron Hoffman; Patrick Brown; Steven Lowrey. 13a. TYPE OF REPORT 13b. TIME COVERED 14. DATE OF REPORT (Year, Month, Day) 115. PAGE COUNT Technical I FROM RR Noy TOB&_fa December I 16. SUPPLEMENTARY NOTATION 17. COSATI CODES 18. SUBJECT TERMS (Continue on reverse if necessary and identify by block number) FIELD GROUP SUB-GROUP - 2 Hexachloroethane -...,. High Pressure Liquid Chromatography I I White Smoke Mixes 19. ABSTRACT (Continue on reverse if necessary and identify by block number) An HPLC method for analysis of hexachloroethane purity and its percent by weight in smoke mixes was investigated, developed and tested. The HPLC method has been shown to be faster, more efficient and safer. A precision and accuracy study indicates the method to be accurate and reliable. The method involves analysis in acetonitrile solution on a reverse phase C-18 column using a UV detector set at 215 nm.,, DISTRIBUTION/AVAILABILITY OF ABSTRACT 21 ABSTRACT SECURITY CLASSIFICATION 3UNCLASSFIED/UNLIMITED 03 SAME AS RPT. 0 DTIC USERS Unclassified 22a. NAME OF RESPONSIBLE INDIVIDUAL 22b. TELEPHONE (Include Area Code) 2Zc. OFFICE SYMBOL Ronald W. Wise SMCPB-QAL-A DO Form 1473, JUN 86 Previous editions are obsolete. SECURITY CLASSIFICATION OF THIS PAGE

3 DISCLAIMER The findings in this report are not to be construed as an official Department of the Army position unless so designated by other authorizing documents. DISTRIBUTION STATET Approved for public release; distribution unlimited. Acce-qr,, F~r NTIS fatv DTIC TAB C] By Dist, ibtijo I Avdi.Jbi,ty Codes Dist D Avjil andj/ or Spo'cial_ L I " I II I II I!i

4 CONTENTS 1. Introduction Experimental Procedures Materials Gas Chromatography/Mass Spectrometry High Pressure Liquid Chromatography (HPLC) Smoke Mix Controls for Suitability & Precision & Accuracy Studies_ Precision and Accuracy Study Design Data Analysis Results and Discussion Mobile Phase Conditions Wavelength Selection Hexachloroethane Standard Purity Identity of Standard and Hexachloroethane in Authentic Smoke Mixes Suitability for Determination of Purity Suitability for Determination of Hexachloroethane in Smoke Mixes Conclusions Literature Cited Appendix A: Statistical Treatment of Calibration Data Appendix B: Statistical Treatment of Data PAGE

5 LIST OF FIGURES PAGE 1. HPLC Chromatogram of Hexachloroethane Hexachloroethane Peak Response at Selected Wavelengths Mass Spectrum of Hexachloroethane Analysis of Precision and Accuracy of the Method Results of Injecting 4 Concentrations in Quadruplicate Calibration Curve for Day 1, Smoke Mix P & A Study Calibration Curve for Day 2, Smoke Mix P & A Study Precision and Accuracy for Hexachloroethane One-Way Analysis of Variance LIST OF TABLES 1. Comparison of Peak Areas Obtained for Hexachloroethane Using 2 different Mobile Phases and 3 Different Wavelengths P & A Study for Hexachloroethane in Smoke Mixes Calculated Quantities of Hexachloroethane in Smoke Mixes

6 HEXACHLOROETHANE PURITY AND ASSAY IN SMGKE MIX BY HIGH PRESSURE LIQUID CHROMATOGRAPHY MTT TASK # INTRODUCTION Hexachloroethane is a component of white smoke mixtures used by the Department of Defense. Pure hexachloroethane is currently purchased using specifications set forth in MIL-H-235C dtd 23 Feb This specification requires the material to be a minimum percent hexachloroethane by weight. The assessment of purity (para ) requires the use of procedures in both ASTM E256 and ASTM D2989. These procedures require ignition in a sodium peroxide bomb which is hazardous and labor intensive. For smoke mixtures, the percent by weight of hexachloroethane is currently determined by heating the mixture in a vented oven for two hours at 200*C. In order for the procedure to be valid, water content must also be determined using ASTM E203 with Karl Fischer Reagent and hexachloroethane must be the only volatile component besides water. Thus, determining the percent by weight of hexachloroethane in smoke mixes is currently a multistep process of questionable reliability which generates hazardous waste and fumes. A faster, more efficient and safer method of assessing hexachloroethane purity and percent by weight in smoke mixes was investigated using high pressure liquid chromatography (HPLC). 2. EXPERIMENTAL PROCEDURES 2.1 MATERIALS Hexachloroethane was provided by the Pine Bluff Arsenal. Standard material was prepared and purity assessed as indicated in the results section. All solvents were purchased as HPLC grade. Smoke mix components Zinc Oxide and Aluminum powder were obtained from the stocks actually being used on the production lines.

7 2.2 GAS CHROMATOGRAPHY/MASS SPECTROMETRY Hexachloroethane standard was analyzed on a Finnigan Model 5100 GC/MS using electron impact ionization. The oven fitted with a 30m DB1701 capillary column was programmed from 30C to 90oC at 10*C/min, held at 90C for 5 min, then increased to 230 C at 20 C/min. Injection sample size was 1 ul, injector temperature was 200*C, interface temperature was 230 C, and the ion source temperature was C. The mass range was scanned from amu at two scans per second. Calibration and tuning was with FC-43, perfluorotributylamine. 2.3 HIGH PRESSURE LIQUID CHROMATOGRAPHY (HPLC) HPLC analysis utilized a mobile phase consisting of acetonitrile/water (75/25) flowing through a Brownlee Spheri-5 reverse phase 5 um, 250 x 4.6 mm column at 1 ml/min. The detector was an LKB 2151 set at 215 nm. Injections were made with a Rheodyne 7125 valve fitted with a 50 ul loop. The loop was loaded using 100 ul of sample. Results of each injection were recorded on an LKB 2220 recording integrator. Areas under each peak were obtained using an attenuation of 2 x 106, chart speed 1 cm/min and a threshhold of 2 x SMOKE MIX CONTROLS FOR SUITABILITY AND PRECISION AND ACCURACY STUDIES Smoke mixes of known composition were difficult to prepare because of the volatility of hexachloroethane and difficulty getting a uniform blend of the components. Suitable mixtures were finally achieved by mixing the solid using a 3/8 inch stir bar in a 4 ml closed container filled to 80% capacity and placed over a magnetic stir plate. All weighing of solid samples was achieved as quickly as possible. QP's for the P&A study were prepared by weighing the smoke mix components directly into acetonitrile. 2.5 PRECISION AND ACCURACY STUDY DESIGN A set of calibration standards was prepared at concentrations from 17.5 to 32.5 ug hexachloroethane per 50 ul acetonitrile in 100 ml volumetric flasks. A set of artificial smoke mixes (QP's) was prepared at 3 concentrations in quadruplicate representing smoke mix compositions of 40, 50 and 60% hexachloroethane. 2

8 A 25 ug per 50 ul standard was used as a QL to make sure the instrument was not drifing out of calibration. All samples were injected in the order presented in Table 2 on two consecutive days. Peak response is peak area of a single injection as recorded by the LKB integrator. 2.6 DATA ANALYSIS Statistical analysis of calibration data was by a least mean square fit to a linear equation as described in Appendix A. Statistical treatment of the data from the Precision and Accuracy Study of smoke mix analysis is described in Appendix B. The detection limit and confidence limits are determined. 3. RESULTS AND DISCUSSION 3.1 MOBILE PHASE CONDITIONS HPLC conditions were chosen which gave a retention time of about 8.2 minutes. The standard hexachloroethane sample was found to have an impurity that was not adequately resolved at shorter retention times and completely hidden at the 3.6 minutes retention time which resulted from using a 100% acetonitrile mobile phase. An acetonitrile/water (75:25) mobile phase flowing at 1 ml per minute resulted in a symmetrical peak with no apparent hidden impurities (Figure 1). The lack of unresolved impurities was indicated by the unchanging ratio of peak areas obtained at three different wavelengths and two different mobile phases (Table 1). The 13.6 minute retention time peaks were obtained using an acetonitrile/water (65:35) mobile phase flowing at 1 ml per minute. 3.2 WAVELENGTH SELECTION Hexachloroethane did not exhibit a characteristic lambda max in the range 190 to 600 nm range. A wavelength was chosen, therefore, which maximized the signal to noise and reproducibility. As the wavelength is decreased, sample absorbance increases as well as background noise. As the wavelength is increased, sample absorbance decreased and electronic amplification noise necessarily increases. The best wavelength may have to be chosen for each detector depending on its level of sophis- 3

9 tication. The longer wavelengths are less likely to have interference from contaminants. Figure 2 shows the results of injecting the same amount of hexachloroethane in acetonitrile with the detector set to various wavelengths. 3.3 HEXACHLOROETHANE STANDARD PURITY The purity of the standard provided by the Pine Bluff Arsenal was determined to be 89.3% by weight. Purity was determined by comparison of HPLC peak heights with a sample prepared by sublimation on a cold finger. Less than 1% contamination of the sublimed material could be detected by HPLC or GC/MS. The melting point of sublimed material was C while that of the PBA standard was 185.5C. By comparison, a certified 98.7% purity sample (Ref. 7) had a melting point of 185.6*C. The volatility of the sublimed material at 200C for one hour was 100% while that of the PBA standard was 99.3%. The identity of the major contaminant(s) is currently under investigation. At least one contaminant is readily trapped by the sublimed crystals and has very similar HPLC chromatographic characteristics as hexachloroethane. GC/MS analysis of the PBA standard identified a small amount of tetrachloroethane eluting before hexachloroethane. 3.4 IDENTITY OF STANDARD AND HEXACHLOROETHANE IN AUTHENTIC SMOKE MIXES The mass spectrum shown in figure 3 indicates the presence of a hydrocarbon containing six chlorine atoms, a molecular weight of 234 and matches the spectral library for this compound. The material assessed in smoke mix as hexachloroethane had the same retention times on HPLC and GC and the same mass spectral pattern as authentic hexachloroethane (Ref. 6). 3.5 SUITABILITY OF HPLC METHOD FOR DETERMINATION OF HEXACHLOROETHANE PURITY Eighteen calibration standards were prepared at concentrations from 5 to 50 ug per 50 ul acetonitrile and injected into the HPLC in triplicate. Using the average peak response for all 36 injections as a calibration factor, the ug content of each sample was calculated and plotted versus its actual concentration 4

10 (Figure 4). A Hubaux and Vos analysis of this data indicates a lower detection limit for 2 replications to be 2.87 ug with a variance of The correlation coefficient for this data was Additional evaluation of the method was performed by preparing standard hexachloroethane at 4 different concentrations and injecting each 4 times. A least squares fit was performed on the data generated. The correlation coefficient of the relationship between concentration and peak area was found to be (Figure 5). The following peak areas were obtained for 4 injections of each of the following standards: lo.02ug/looul 20.13ug/lOOul 50.55ug/lOOul ug/lOOul X 1,640,825 3,191,450 7,957,875 15,899,250 S.D. ±0.588% ±1.34% ±0.565% ±0.529% Analyzing the response factors (peak area/ug injected) for each injection of the 4 different standards yields: Response Factor x = 316,632 (n=16) S.D. = ±2.49% Using this Response Factor and the average peak areas for the different standards, a simple back calculation yields the ug injected, the concentration of standard solution and hence, the percent purity: 10.02ug/100ul 20.13ug/100ul 50.55ug/lOOul ug/100ul ug inj conc % pur SUITABILITY FOR DETERMINATION OF HEXACHLOROETHANE IN SMOKE MIXES A set of calibration standards was prepared at concentrations of 17.5 to 32.5 ug per 50 ul acetonitrile. A set of artificial smoke mixes (QP's) was prepared at 3 concentrations in quadruplicate representing smoke mix compositions of 40, 50 and 60% hexachloroethane. A 25 ug/50 ul standard was used as a QL to make sure the instrument was not drifting out of calibration. All samples were injected once per day for two days. Peak areas and retention times are shown in Table 2. A least squares fit program was used to determine a linear relationship 5

11 between the micrograms of hexachloroethane in the standards to their respective peak areas for each of the twc day's trials (Figure 6 and 7). From this relationship, the instrumenital found micrograms of hexachloroethane present in the QP samples and QL samples analyzed on each day were calculated (Table 3). The QL samples were analyzed for percent imprecision and percent inaccuracy at the 95% confidence level for each day, and for the total analysis. % inaccuracy % imprecision Day #1 QL 0.43% ± 0.70% Day #2 QL % ± 0.71% Total QL 0.16% ± 0.52% The QP samples were adjusted to conform to the analysis being carried out using three select quantities of hexachloroethane injected. This was accomplished by ratioing the select concentration to the actual quantity injected and by comparing direct>y this proportion to the ratio of the calculated recovered amount of hexachloroethane to the known amount of hexachloroethane recovered. i.e. for QPG day #1: 20 ug - x ug ug where x is the calculated recovered amount. The select quantities of hexachloroethane injected and the calculated recovered amounts were determined to be as follows: SELECT QUANTITIES 20ug 25.05ug 29.7ug Day # Day # The calculated recovered amounts were analyzed using a basic program based on the methods of Hubaux and Vos to determine a lower detection limit (Figure 8,. 6

12 This was found to be 1.03 ug of hexachloroethane. This data was also subjected to a statistical analysis to determine the percent impreci:ion, and percent inaccuracy at the 95% confidence level for each select quantity of hexachloroethane injected (Figure 9). select quantity tnjected % inaccuracy % imprecision 20 ug % ± 0.67% ug % ± 0.53% 29.7 ug % ± 0.45% 4. CONCLUSIONS This HPLC method is a precise and accurate method for determination of hexachloroethane in pure sample purchases and in white smoke mixes. The use of this method should increase the level of confidence in these determinations over the old methods. The method is less susceptable to operator error and much safer than the sodium peroxide bomb method. Hazardous substances produced in this testing should be less hazardous to laboratory personnel and the environment. 7

13 ACKNOWLEDGEMENT The authors wish to thank Joanie Payne for expert secretarial assistance. LITERATURE CITED 1. MIL-H-235C, 23 Feb 1984, Military Specification for Hexachloroethane, Technical, published by DRSMC-TSC-S(A), Aberdeen Proving Ground, MD ASTM E256-67; Standard Method of Test for "Chlorine in Organic Comounds by Sodium Peroxide Bomb Ignition", American National Standard K , American Society for Testing and Materials, 1916 Race St., Philadelphia 3, PA. 3. ASTM E203-64; Standard Method of Test for "Water Using Karl Fischer Reagent", American National Standard K , Reapproved 1971, American Society for Testing Materials, 1916 Race St., Philadelphia 3, PA. 4. ASTM D2989; Standard Method of Test for "Acidity - Alkalinity of Halogenated Organic Solvents and Their Admixtures", American Society for Testing & Materials, 1916 Race St., Philadelphia 3, PA. 5. Hubaux, A and Vos, Gilbert, 1970 "Decision and Detection Limits for Linear Calibration Curves", Analytical Chemistry, Vol. 42, No. 8, pg NBS Library Compilation, Entry No , CAS No , Finnigan Corp., San Jose, CA P.J. Heinrici, Inc. "Sampling and Analysis of Hexachloroethane at American Warehouse, Houston Texas on March 16, 1987 and Certificate of Analysis 3113 Red Bluff Road, Pasadena, TX

14 FIGURE 1 HPLC Chromatogram of Hexachloroethane Standard Hexachloroethane WImpurity 0 I Time (min) I Mobile phase was acetonitrile/water (3:1) flowing at 1 ml/min through a 25 cm C-18 RP column. LKB 2220 inteqrator was set with ATT2t =6. PK. WD = 0.04 and THRSH=4. 9

15 FIGURE 2 A plot of Hexachloroethane peak areas obtained at various wavelengths by injecting 19.6 ug standard using a mobile phase of acetonitrile/water (3:1) xa 0 U - 0 a, S 4 0U p4 2' 01

16 I-X K' in M- I II LD-m M 1 1+ 'A 1 =.f.1 m 11 1"

17 F I GU RE 4 Analysis of Precision and Accuracy of the Method Y + t SFIKED wit 3f 4EX (ug) HEXACHLOROETHANE, SPIKED vs FOUND 12

18 FIGURE 5 Results of Injecting 4 Concentrations of Hexachloroethane in Quadruplicate y : A*% + B A= , B= Correlation= Hexachloroethane Mix a s P e a o Concentration (magml) 13

19 7 6 5! 6 ~~~~ ~ 3MOOU 25. ~ 913 ~ T'0" c. E I. UNE~."T"I ' f q C~' C L R O ~ H N D... A B L.E.. A N K X L..... CaVrto uv rdy1 mk i td.~~... ~... ~. F.. G... R

20 6!000O ) 2O3.400,00 1I s 1 1a A HEXCHORETHNEMT, AY#,BAKEC Caliratin Crve or a-,soemx td F A A15

21 PRECISION AND ACCURACY FOR HEXACHLOROETHANE TARGET INPUT AVERAGE STD PERCENT PERCENT VALUE VALUES DEV COEFF INACCURACY ug VAR LEAST SQUARES FIT *** Y INTERCEPT = VARIANCE = E-04 SLOPE = FOR 1 REPLICATIONS LOWER DETECT LIMIT X(D)= WITH Y INTERCEPT Y(D)= FIGURE 8 16

22 FIGURE 9 One-Way Analysis of Variance Data: known Level codes: spike Labels: Range test: Conf. Int. Confidence level: 95 Analysis of variance Source of variation Sum of Squares d.f. Mean square F-ratio Sig. level Between groups Within groups Total (corrected) missing value(s) have been excluded. Table of means for known by spike Stnd. Error Stnd. Error 95 Percent Confidence Level Count Average (internal) (pooled s) intervals for mean ± Total Percent Confidence Intervals for Factor Means 29 r k o 25.1 w n 2 3 E w=.f= level of spike 17

23 Regression Analysis - Linear model: Y = a+bx F I G U R E 9 Continued Dependent variable: known Independent variable: spike Standard T Prob. Parameter Estimate Error Value Level Intercept ±008 Slope i4714E Analysis of Variance Source Sum of Squares Df Mean Square F-Ratio Prob. Level Model Error.693: Total (Corr.) Correlation Coefficient = R-squared percent Stnd. Error of Est Regression of known on spike ", w o IIIIII spike 18

24 TABLE 1 Comparison of peak areas obtained for hexachloroethane using 2 different mobile phases and 3 different wavelengths. RETENTION TIMES WAVELENGTH 8.4 minutes 13.6 minutes (nm) Area x 10 7 S.D. Area x 10 7 S.D RATIOS 210/ a / a. Not significantly different from the ratio at 8.4 minutes at the 95% confidence level. 19

25 TABLE 2 P&A STUDY FOR HEXACHLOROETHANE ASSAY IN SMOKE MIXES PEAK RESPONSE Order of Quantity Day Residence Day Residence Injection Injected (ug) I Time (min) 2 Time (min) (XIO 6 ) (XO ) Std A is B ap C ,, D ,a E aa F QP G a H is I QL C QP J % oa K to L QL C QP M a' N of F QL C QP P af Q at R QL C

26 TABLE 3 CALCULATED QUANTITIES OF HEXACHLOROETHANE IN SMOKE MIXES SAMPLE ID QTY INJECTED (ug) DAY 1 CALCULATED (ug) DAY 2 CALCULATED (ug) QP G to A a QL C QP J QP K It L QL C QP M ia N i QL C QP P QP Q QP R QL C

27 APPENDIX A Statistical Analysis of Calibration Data 1. Assume the calibration curve is linear, and can be described by the equation: Y = Yo+bX b = N(sumXiYi) - (sumxi) (sumyi) N(sumXiA2) - (sumxi)a2 Yo = (sumyi) - b(sumxi) N Source of Sum of Degrees of Mean Square (MS) Variation Squares (SS) Freedom (df) Residual 0 N-2 (ResidualSS) / (N-2) Total Error (sum d^2)/2 N-2 (Total ErrorSS)/ df Total Error Lack of Fit Residual SS dfresidual LOFSS/df LOF (LOF) -Total SS -dftotal F-Ratio=(MS LOF)/MS Total Error Q=((sumYi^2)-(sumYi)-2/N)-b 2((sumXi 2)-(sumXi)a2/N) N = number of data points Xi= i-th target concentration Yi=!-th value of dependent variable d = difference between duplicates 22

28 APPENDIX B STATISTICAL TREATMENT OF DATA The detection limit and confidence limits will be determined in either of two acceptable approaches. The first approach is to use a computer program that the Product Assurance Laboratory possesses that utilizes the mathematical treatment of Hubaux and Vos (Ref. 5). The second approach is described as follows. For Y=Yo+bX b = N(sumXiYi) - (sumxi) (sumyi) Yo = (sumyi) - b(sum)i) N(sumXi^2) - (sumxi)^2 N Upper confidence limit = Yu Yu = Yo+bX+Sx.yT(I+l+(Xi-X" )"2 )"(1/2) ( R sum(xi-x')^2) Lower confidence limit = Yl Where: Yl = Yo+bX-Sx.yT(l+l+(Xi-X ) ^2 )^(1/2) ( NT sum(xi-x")a2 ) Sx.y = (sum(yi-(y +b(xi-x ) A (1/2) N-2 T = student's T for 2-tailed P=O.10 and N-2 degrees of freedom N = number of data points Xi= i-th target concentration Yi- i-th value of dependent variable X'= mean value of Xi's Y'= mean value of Yi's The calculated reporting limit is determined by drawing a horizontal line from the Y-intercept of the upper confidence curve to its corresponding value on the lower confidence curve and reading the X value fo' this point on the lower confidence curve. This value is the certified reporting limit as long as one of the tested concentrations was at or below this value, otherwise, the lowest tested value is the certified reporting limit. 23

29 APPENDIX B Continued The slope of the least squares linear regression line of a plot of found versus target concentrations is a measure of the accuracy of the method. Standard deviation = S S = (sum(yi"2) - ((sumyi)a2)/n)l/2 C N-I Percent inaccuracy = YtOX (100) X X = target concentration Yt'= average found concentration at the target concentration Percent imprecision = S,(100) Yt 24